Trends in practice patterns at Duke University Medical Center were assessed in patient groups comparable to those enrolled in the three major randomized trials of coronary artery bypass grafting (CABG). In addition, changes in practice patterns that appeared unrelated to the randomized trials were examined. Most patients with 75% or greater left main stenosis have been treated surgically after publication of the Veterans Administration Cooperative Study, but little change was noted in the proportion of patients with 50% left main stenosis who have been treated surgically. A trend towards selection of surgical therapy for patients with three-vessel disease and normal left ventricular function was evident before the publication of the European Coronary Surgery Study, although one-third of patients in this category continue to be treated nonsurgically after publication of the results of the trial. For the past decade, most patients who would have been eligible for the Coronary Artery Surgery Study have been treated nonsurgically. We have also documented trends in practice patterns that are independent of the results of randomized trials. The advent of percutaneous transluminal angioplasty has provided another therapeutic alternative that has been used increasingly. In addition, growing numbers of patients with advanced age, unstable angina, or markedly depressed left ventricular ejection fraction are being evaluated with cardiac catheterization despite the lack of supporting randomized trials. Randomized trials have placed our understanding of the effects of CABG on a sound foundation, but it is evident that clinicians continue to consider many other factors when therapeutic decisions are made.